Tube emulation with FETs

[rubberlips]

G'day guys,

I was wondering how to go about converting a tube circuit into a circuit with FETs.  Whats involved with resistor and cap values?? Is it just a matter of changing the tube pins for the equvalent FET pins?? Obviously the rail voltages would be different but other than that.?

Is there some kind of tutorial on the subject??

Cheers

Pete

[JimRayden]

Thre are some minor component additions and changes but the main circuit remains the same, it just runs on 9 volts and doesn't need a filament supply. Compare the ROG Professor Tweed to a Princeton scheme. No much difference, is there?


-----------------
Jimbo

[rubberlips]

Thanks jim
no, not much difference at all.  I was going to run it more at about +/-15V instead as a preamp, would that make a difference?

Cheers

Pete

[JimRayden]

Check your transistors' voltage handling.

Or, I'll let someone wiser handle that question.

------------
Jimbo

[stm]

You can find a guideline in the text of this article:

http://www.runoffgroove.com/professor.html

Also, check the third page of this thread, where the origins and evolution of FET emulation is covered:

http://www.diystompboxes.com/sboxforum/viewtopic.php?t=30093&start=30

Using +/- 15V won't produce the same result.  TUBE amps used a single supply for all its pre and power stages.  You can still have dual supplies to feed opamps and other stuff, but feed the JFET stages from a single supply only. Also, maybe 15V is too high to get clipping without lots of gain.  You can try having a dedicated 9V or 12V supply for the JFET stages.

Regards,

STM

[rubberlips]

You can find a guideline in the text of this article:

http://www.runoffgroove.com/professor.html

Also, check the third page of this thread, where the origins and evolution of FET emulation is covered:

http://www.diystompboxes.com/sboxforum/viewtopic.php?t=30093&start=30

Using +/- 15V won't produce the same result.  TUBE amps used a single supply for all its pre and power stages.  You can still have dual supplies to feed opamps and other stuff, but feed the JFET stages from a single supply only. Also, maybe 15V is too high to get clipping without lots of gain.  You can try having a dedicated 9V or 12V supply for the JFET stages.

Regards,

STM

Thanks I'll keep that in mind. The reason I was going to use +/-15V was that most SS amp power supplies usually have either +/-15 or +/-12V outputs. I suppose I could change the drain resistors to give less voltage.

Plus I suppose you'd have to bias the FETs differently to go into an amp with +/- supply rails?

Cheers

Pete

[JimRayden]

If you have +/-12V, then use +12V and ground for the circuit. Leave the negativepart of the supply out. Simple as that.


---------------
Jimbo

[Transmogrifox]

I didn't read the articles about the ROGS, they may have covered this, but here's something you need to take into consideration:

headroom to gain ratio.

For example, a 12AX7 has a typical gain of 100.  Suppose we have a circuit where a 1V amplitude signal is at the input.  In this case, the output will want to swing 100V.  Suppose we have a lower headroom on the 12AX7 stage of 75 volts.
The headroom to gain ratio in this case is 0.75.

In this case, it takes a 750mV input signal to cause the output to clip.

In a JFET stage with the same gain, you have a headroom of say, 10V.  therefore it takes a 100mV signal to drive it to the rail.  The headroom to gain ratio in this case is 10/100 = 0.1.  As you can see, you need to do some gain adjustment as well as correct biasing to ensure that the headroom to gain ratio is equivalent to the tube stage for both polarities.  It's not enough to simply assume the tube stage is always symmetrical, so you must do this for both the cut-off and saturation case.

This is one of the reasons why you can't just use the same bias resistors and bypass caps, merely replacing the tube with a FET to accurately model the amp.

[petemoore]

I've been using low voltage circuits. Jfets work great in one I have...2 or 3 stages @ ~3V.
 The other is AMZ's Si into Jfet @ 1.5V ... very soft Distortion.

[MR COFFEE]

Too much misinformation showing up here.

1) Higher voltage operation is not prohibited with many JFETs. It can help get the ~ 50:1 clipping ratio for clipped tube emulation. Look up the Jack Sondermeyer (Peavey) patents for cascaded tube stage emulation if you want to read a tutorial on how this aspect of distortion works. As noted above, a diode can cover the grid conduction and bias shift issue quite effectively.

2) Low voltage operation is good for clipping if assymetry is not the object. The BOSS OD-2 runs differential pairs in a discrete feedback "op amp" configuration on 4.5 volts.

3) I am NOT slagging the runoff groove "amp emulators" here, but another issue not being noted by many here searching to emulate the overdriven tube sound is that the frequency and time-constants used in a tube's Cathode bias circuit cannot be duplicated by putting the same values in the Source bias circuit of a JFET. The capacitors' effect on the low frequency roll-off is related to the transconductance of the tube or JFET stage, *NOT* the RC time constant of the network. Read Aiken amp tutorials if you want to understand this relationship in greater detail in order to emulate it.

4) MANY JFETs will run fine on 30 volts, even 40 or more, and running JFET single-ended circuits from quiet +/- 15volts rails works just fine. Just decouple your bias circuits if you are deriving your +/- 15volts from a switching converter module instead of a nice quiet little analog supply. And filter the rails with a small pi filter as well as some good Low ESR reservoir caps of 100uf or so, isolated from the converter module with 25-100 ohms.

5) Remember, the VOX AC-30 emulator was made to emulate the tube characteristics using op amps, diodes, and other clipping devices with resistors in series to give the classic piece-wise approximation to the tube circuit overdrive characteristics. But the circuit values were derived by MEASURING the VOX AC-30 circuit one stage at a time and studing the transfer function of each stage before selecting values to use in a circuit. GOOD EMULATION IS WORK.
An oscilliscope is really important if you want to get somewhere at this business. So is a source of a string-like stimulus wave (I prefer a fast-attack, exponential-decay triangle wave generator, just so I can see the hard and soft clipping thresholds more readily, but that's just me). Some folks use fast-attack, exponential-(or even linear)-decay sine waves and have awesome results. Of course, you can learn quite a bit watching while you play through a circuit on a scope and twiddling things.

FWIW.

Have fun. But don't think the road to JFET tube nirvana is that easy. If you just want to copy somebody's circuit, get a copy of the Roland BC-60 and start from their preamp circuits. Then start listening and watching that scope and see what's going on and what it sounds like. And start tweaking.

Enjoy,

[amz-fx]

An excellent post Mr Coffee and right on target!  

One of the big limits of getting a good amp emulation is the 9v battery...  my Mini-Tubes circuit was designed for 30v - 35v, but would work on 9v with lesser results. Anyone interested in amp emulators should search the ARCHIVES and FIRST STOMPBOX FORUM sections of this message board and read some of the threads.  You will not only see posts about the Mini-Tubes but also how the other derivatives such as the Booster 2.5, BSIAB, Sweet Thing and others came to be.  There is lots of good information that we were kicking around 5 - 6 years ago!  Just a sample:

http://diystompboxes.com/sboxforum/viewtopic.php?t=15431
http://diystompboxes.com/sboxforum/viewtopic.php?t=4107
http://diystompboxes.com/sboxforum/viewtopic.php?t=15033

and also: http://www.muzique.com/schem/Aron_archives13.html

The point you make about "frequency and time-constants" is well taken and something I learned through a lot of trial and error...  while I was researching jfet circuits, I built an amp emulator stage using a J201, connected it to the scope and coincidentally it had a rolloff at around 2.2k Hz just as I had planned.  I dropped in a 2N5484 and readjusted the bias only to find out that the corner frequency had moved to over 3K Hz! What the?!? Merely changing transistors had changed the frequency response by almost 1k Hz.

This brings up another  point...  even if you go to the trouble to tweak and tune a circuit to have exactly the same frequency response as the tube amp, that response will change when you make the next copy unless the jfet has exactly the same characteristics of the one you used to develop the circuit...  this means: no trimmer on the drain!  You must get the frequency response worked out and then select jfets that exactly match the original...  i.e., buy 100, sort and use 10 or so.

BTW, I usually use a static triangle wave of about 200Hz when doing preliminary testing because it is easy to see the top or bottom "round-over" as the circuit begins to clip.

Here is where I'm at these days...  http://209.124.248.167/media/misc/tweed.jpg  This is the outside and inside of my Tweed Stack pedal. Notice there is no battery...  it runs off a custom 36v power supply. There is a special preamp chip on the input and a jfet somewhere there in the middle.  This comes much closer to emulating an amp than simply putting a bunch of jfet stages in series...  FBIAB (Fender Bassman in a Box!)  

best regards, Jack

[R.G.]

Let me just reiterate some of this and restate what I say every time this comes up:

JFETs are not tubes.
(1) Although depletion mode JFETs can be biased like triodes, they do not have the same characteristics. A whole world of the ideas about high voltage, biasing, signal-to-power-supply, and other considerations are subsumed here.
(2) JFETs are square law devices, tubes are 3/2 power devices so their non-clipped distortion is fundamentally not the same. In many ways, I prefer the square law device distortion, but it's not the same. Bipolars are, by the way, exponential law devices and fundamentally different yet.
(3) JFETs have massive variations from device to device that are missing in tubes, which are really quite consistent within a type number.
(4) The quest for JFET emulation of tubes says more about human psychology than it does about JFETs. The cachet of "sounds like tubes" is practically overwhelming in the music world, I guess. It makes great advertising copy, and it's what buyers want to hear.
(5) If JFETs could emulate tubes closely, then far better and more aggressive minds that we have here would have achieved it as a commercial product in the forty years that we (humans) have had JFETs. It is not the case that EE/musicians (and there are really quite a lot of those) were just too stupid or lazy to see the "obvious". Further, the economic potential to manufacturers to do this is just too great for that not to have already happened.

That is not to say that JFET devices can't sound good. I suspect that if the devices did not vary so much, they would have supplanted tubes anyway for that reason. Designing out sensitivity to variation is a bedrock of bipolar design practice, and the same things can be applied to JFETs. But if you do that, you design out the things that make them sound unique, because you design for the device not mattering.

And simply replacing triodes with JFETs as has already been noted will not, in general, sound like a tube circuit.

So enough with the "JFETs emulating tubes" already.  JFETs sound good in their own right (just as opamps do, and bipolars and...) when designed properly. So use them where  they sound good and enjoy them for that. But it's not very productive to assume that if you just easter-egg in one more tweak you're going to find your own beautiful reality where hollow-state devices are parenthetical. Use JFETs because you like them, not because if you change ...one ... more... component... they'll become a non-heater version of a 12AX7. Muck with them and say "Hey! That's a really great JFET sound!"

Love them for what they are, don't date them for what you want them to be.

[maximee]

hey jack,
I'm pretty curious about your pedal...do you have sound clips?  

[javacody]

Well, with all of our technology, where are the low-voltage tubes that last forever? We can send people into space, but we cannot accurately model a tube?

[R.G.]

Well, with all of our technology, where are the low-voltage tubes that last forever?

They are available. They're called Field Emission Triodes, sadly the same acronym as Field Effect Transistor. Field Emission Triodes use an array of micromachined needle points on a silicon surface to emit electrons with low voltages. Put two conductive rings around a pit full of these needles and enclose it in a vacuum, and you have a triode. In fact, you have a planar process, photolithography built, integratable triode technology that is radiation hard.

Why can't you buy them? Because they are economically infeasible. No one will risk the million or ten to build a batch and see if you come to buy them. They are good for military logic in high radiation and space shot environments, but there's simply not a market dependable enough to make them marketable. And they're special purpose devices, not individual triodes.

Engineering economics is a hard master.

We can send people into space, but we cannot accurately model a tube?

We can quite accurately model tubes. We just can't economically model tubes in a way that the guitar amp people like. Like it or not, Mother Nature is an economist by trade.

And by the way, we went to the moon when I graduated high school. Then we came home and for the last three decades never went back. That's an economic result: we (the people) are in general not willing to pay for it. It's not that we can't. We refuse to pay for it. You personally might think it's great and be willing to pay for it, but you have to have 100M people agreeing at the same time to get up the collective will to do it.

[javacody]

It's so easy to forget about the almighty dollar when you are just approaching things from a hobbyiest perspective. Thanks for the reminder.  

[Arno van der Heijden]

I came across this interesting tube emulator, developed by Eric Pritchard:
http://www.pritchardamps.com/pritchardamps/tech_talk/db_magazine.pdf

Anybody ever tried something like this?

[puretube]

tnx for that link, Arno!!!

[puretube]

it seems not as easy,as in that article, though... :
http://v3.espacenet.com/textdraw?CY=ep&LG=de&F=4&IDX=US5434536&DB=EPODOC&QPN=US5434536
maybe Joe Davisson is on the tracks, with his "diode compression"?

[Arno van der Heijden]

puretube, is there a way to save the whole patent in one pdf file? Right now, I have to save every single page  :evil:

Btw, here's a list with some more interesting(?) patents:
http://www.pritchardamps.com/pritchardamps/tech_talk/patents.htm